AI Agent Engineer | Multi-Agent Systems, RAG & LLM Orchestration

Building autonomous AI systems that work in production — not just demos.

I specialize in AI agent development and multi-agent orchestration. The kind of systems where an AI doesn't just chat — it reasons, retrieves verified information, collaborates with other agents, and takes action autonomously.

Current Focus:

• Multi-agent systems with LangGraph + DSPy
• Production RAG pipelines with citations & audit trails
• Agent orchestration infrastructure (MCP protocol)
• Business automation with AI agents

The flagship project. Three AI agents (2 Examiners + 1 Arbiter) grade complex essay questions autonomously. When agents disagree beyond a threshold, the Arbiter reviews both arguments and makes the final call — mirroring real academic committees.

• Results: 5x throughput improvement, 90% fewer DB queries via intelligent caching
• Tech: LangGraph, DSPy, LangChain, ChromaDB, Streamlit
• Status: In pilot with professors at Instituto Federal Fluminense

In 2026, teams use multiple AI coding agents (Claude Code, Antigravity, Cursor) — but they don't talk to each other. Agentbridge solves this.

Define workflows in YAML. Each step is assigned to a specific agent. When one finishes, the next unblocks automatically. Dependency resolution, conditional execution, cross-agent handoff — all handled.

• Verified E2E with 2 simulated agents via HTTP sessions
• Tech: TypeScript, MCP (Model Context Protocol), SQLite, YAML
• Differentiator: One of the first MCP orchestrators on GitHub

Autonomous AI agent that aggregates metrics from Shopify, Meta Ads, and Google Ads, detects anomalies against rolling baselines, and generates prioritized action items daily.

• What it solves: Reduces 45-60 min/day of manual dashboard checking to seconds
• Tech: FastAPI, SQLite, Docker, Jinja2 templating
• Architecture: Clean adapter pattern, decision memory, one-click Render deploy

Turn WhatsApp into a full agentic interface. Send a message from your phone, Claude executes on your Mac with full tool access, and sends back the result — including modified files.

• Features: Streaming with heartbeat, session continuity, auto file sharing, cost tracking, 8 slash commands
• Tech: Python, Claude CLI (stream-json), SQLite, Go (WhatsApp bridge)
• Architecture: Webhook daemon + Go bridge + Claude CLI with process management

Enterprise-ready RAG system with mandatory citations, confidence thresholds ("Not in KB yet" fallback), full audit trail (JSONL + SQLite), and built-in retrieval evaluation harness.

• Tech: LangChain, ChromaDB, OpenAI/Gemini, Streamlit
• Features: Manifest-driven ingestion, recall@k metrics, section-level citations

AI/ML: Python LangChain LangGraph DSPy OpenAI API Anthropic Claude ChromaDB RAG Multi-Agent Systems

Backend & Infra: FastAPI TypeScript Node.js SQLite Docker Git CI/CD

Agent Orchestration: MCP (Model Context Protocol) State Machines Workflow Engines

Act 1 — Foundations (2023-2024) Started with ML fundamentals: implemented PCA from scratch, built neural networks for self-driving car simulation. Understood the math before jumping to LLMs.

Act 2 — RAG Mastery (2025) Dove deep into RAG when I realized most businesses don't need to train models — they need to connect LLMs to their own data. Built stateful RAG systems with LangGraph that decide autonomously when to retrieve.

Act 3 — Multi-Agent Systems (2025-2026) Evolved from simple agents to multi-agent orchestration. In my capstone project (TCC), built a 3-agent grading system — the same pattern used by Anthropic and Google for complex reasoning tasks.

Act 4 — Agent Infrastructure (2026) Now working on the infrastructure layer: how to orchestrate workflows across agents from different platforms. Created Agentbridge, a cross-IDE orchestrator using the MCP protocol.

Petrobras — AI/ML Engineering Intern Working on AI/ML projects at Brazil's largest company (Fortune 500)

Instituto Federal Fluminense — Computer Engineering Graduating April 2026

🔹 Open to freelance projects: AI agent development, RAG systems, multi-agent workflows, AI automation 🔹 Location: Brazil (GMT-3) | Remote-friendly | US/EU hours available 🔹 Rate: $60-80/hr (mid-level specialist)

📧 Contact me for: AI agent development • RAG pipelines • Multi-agent systems • Business automation